Straight Talk About the Cost of Doing Nothing

What's the cost of disorganized data on Wall Street? A portion of our cover package this month on The Cost of Doing Nothing was inspired by former Tower Group analyst and current SVP of enterprise data management software company Golden Source, Tim Lind, who knows a thing or two about the state of data in major firms. He shared some insightful and irreverent thoughts with us recently on the dangers of not managing data properly.WS&T: You've said that nobody ever talks about the cost of the status quo. What is your thinking on that?

Lind: The biggest question we're asked by potential customers is, what's the business case? You can structure a business case and the assumptions laid in the business case to prove just about anything positive or negative you want. The business case itself is imprecise and overly dependent on metrics, which are unreliable anyway. Whereas the inefficiencies in our day-to-day business seem to be accepted as a reality and a cost of doing business. The business case for doing is scrutinized but the business case for doing nothing seems to be generally accepted. I liken it to having a sucker fish attached to you. It's not a shark, it's not it's going to take a limb. But it's unsightly, it keeps you from going to the beach too often and it's slowly draining you of nutrients. When someone offers to remove the sucker fish, you say how much would that cost, you amortize that cost over a certain number of months and decide you can get used to having the sucker fish on you. It's hard for people to get their head around change, uncertainty and something that's going to have a long-term payback without a safety net of rock solid metrics, based purely on business judgment and a vision of a better way of doing something. These are very challenging projects to get approved.

WS&T: What's the worst that could happen?

Lind: I think the worst that can happen is you have an aggregation of old systems on which you bolted new interfaces in an attempt to keep the patient alive -- you've got a 100-year-old patient who probably should have been dead a while ago, you keep the poor thing alive with one heroic intervention after another. Every asset manager we've worked with had a number of different accounting systems or order management systems, in some cases for good reason -- quite often because of acquisitions. They kept things decentralized, independent, off the grid, and allowed each group to fend for itself, rather than go through the more difficult process of consolidating functions and building leverage that would ultimately be consuming. What's the worst that can happen? A gradual drain of nutrients and resources, limited flexibility to deal with change because you have hard-coded solutions, and a larger IT staff to keep these things going.

WS&T: When you talk about aggregating systems from mergers and such, is that just a data challenge, or is that an application integration challenge as well?

Lind: It's definitely more the latter, I think data is only one aspect of that. Data management software abstracts the data away from any one system. A lot of times, an asset manager has multiple accounting systems he or she inherited, and usually one accounting system is used as the security master file, the system of record for securities data. But the system was constructed as an accounting system, not as a security master application, so it's probably limited in terms of the asset classes that it can manage the accounting function for. So to use one system as a system of record and feed other systems with that is a pretty cumbersome process. Our role in helping manage heterogeneous IT environments is to be that system of abstraction, an independent system that can feed all downstream systems independently. We insulate systems from change.

WS&T: Would you call what you have a reference data management system?

Lind: We use the generic term "enterprise data management system." A lot of the sources of content are commercial data vendors -- Bloomberg, Reuters, S&P. Then there's counterparty data; for example, we sell into broker/dealers who use our system to capture data on their institutional customers, to prove the documentation steps -- where's this customer, what type of trading habits do they have, in order to set credit limits. So if you're Sketchy, Inc. in Nigeria your data may yield a different credit rating than a blue chip asset manager located in NYC. There's legal entity information. And there are the positions and transactions that reside in multiple systems that need to be aggregated for the purposes of compliance risk, reporting, and aggregating positions.

WS&T: Can you leave some of the existing databases and systems in place and just have them feed you a duplicate or golden copy, or do you have to make changes to all the existing repositories?

Lind: We're trying to bring some control, governance and visibility to the whole process of capturing, validating and publishing data. It's unrealistic to expect the systems and databases in place today to be replaced any time soon. If you buy the most modern and popular business applications on the buy side right now, say Charles River or Latent Zero, the database is still tightly coupled with the business applications, they're not architected to pull data from remote data stores, so they only function when you feed them data. So we're not trying to replace the downstream systems, we're just trying to feed them with a consistent process and validation of that data. All the data flows through us like a funnel and is subjected to a common set of business rules to ensure integrity and completeness. Then the data is published out to those downstream consuming systems in a consistent way. So the business application doesn't know it's insulated from change, all it gets is a better and consistent source of data. When people use the term golden copy, it usually indicates they're collecting data from multiple sources and comparing those sources against one another as a method to determine what's the most appropriate source and what's bound to be the most accurate source. People can compare a price from two different sources, if that price moved within 5% of yesterday's price, flag it as suspect because it violated a tolerance check, it moved too much. As a secondary validation, let's compare that against another vendor's source. If those prices are within a certain tolerance, OK, the stock moved a lot since yesterday, we consider the price valid. If it fails the first validation check (it moved too much) and then the second, it doesn't match to what other vendors are telling me, I think I've got a suspect piece of data. Then you flag that to a user who researches that exception and makes whatever changes they need to and then publishes that price as the clean, standardized price to every application that requires it.

WS&T: What's that job called?

Lind: Boring.

WS&T: I meant job title.

Lind: I think they're called data management professionals. That's one thing that's changed as well. The stature and skill set of the data management people has increased over the last 10 years. I got my start at a global custodian and was responsible for the global security master file for the company. What I didn't know about international securities was intense. If you looked up operational risk in the dictionary, you saw my picture there. We did stocks and bonds, we didn't know much about derivatives, and we didn't know the 144-a registered share issued in Thailand sold on the secondary market was subject to Reg X, was it the safe harbor provision? There's a lot of information in international securities and the rights of American shareholders that need to be taken into consideration when you build a global security master file. We called ourselves the Australians because we were way down under in the basement of the building and most of my staff came to me because they misbehaved in other groups. We were the penal colony of my company. I thought, maybe you don't want to give me stoners and drunks for employees, because we're going to be exposed to further risk. The operations clerks and trade processing accountants looked down on us. But the risks if we got things wrong were so large that I thought we were not respected for the skills we had. That's changed a lot. If you meet operational data management people today, these people are all smart, a lot smarter than when I was doing my job.

WS&T: They're not in the basement any more?

Lind: A few of them are in the basement.

WS&T: But they're getting more respect.

Lind: Slightly, there's better PR these days. The evolution of the department itself has changed. I think there's a recognition that the function is a profession.

WS&T: A Tower Group study said about 80% of Wall Street firms have a reference data management initiative in place. Does that sound right to you?

Lind: I think there are a lot of different projects. It depends on how you would categorize a reference data project. If you try to build a project that aggregates information from multiple systems in order to do better reporting for management and customers, in a manner of speaking that's a reference data project. I think the question was asked so broadly that it caught a lot of things in that net. We've seen in the last 12 months a significant uptick in interest. I think the buy side will be particularly buoyant in terms of new projects. I think they finally are getting to the point where they're looking at operational leverage and the efficiency in the back office as an important part of their business, whereas before it was all about the rock star portfolio manager and the front office for their models. Asset management is a competitive industry. There are literally thousands of registered investment advisors in the U.S. and between the likes of State Street Global Advisors, Barclays, Fidelity, Vanguard, Oppenheimer, it's a dog fight. The visibility of their expense ratios is an incredibly important attribute now when you select a mutual fund. Ten years ago there wasn't that kind of transparency.

WS&T: Can you think of companies doing a great job, who have really forged ahead with data management?

Lind: There are firms that have winning projects, I'm not sure there are firms that have completed the entire EDM vision -- that's a journey the industry is on.